[12:09] Riddell: I can do that [12:09] Riddell: done [12:10] mjg59: right, you'd need to reboot I'm afraid [12:10] mjg59: I could fix that, but it would break preseeding [12:10] mjg59: alternatively: [12:10] Kamion: Ok [12:10] mjg59: (echo RESET console-setup/layoutcode; echo RESET console-setup/variantcode) | sudo debconf-communicate [12:10] Kamion: I've no problem with rescuing it myself, I just want to make sure it doesn't happen to other people :) [12:10] Riddell: for future reference: https://launchpad.net/people/launchpad-buildd-admins [12:11] ... should kill off console-setup's memory of this === Arador [n=dcg@74.pool80-103-1.dynamic.uni2.es] has joined #ubuntu-devel [12:11] mjg59: if you wouldn't mind then trying it once more, just to confirm before I upload the fix [12:11] Kamion: Still seems broken [12:12] Want debug log again? [12:12] yeah, why not [12:12] thanks mdz [12:14] Kamion: Ok, done [12:14] Kamion: Oh - is there any hope of us getting a font that means I don't have missing glyph boxes in ubiquity? :) [12:14] mjg59: yeah, I've been wondering about that - it's just Dzongkha [12:15] advice gratefully appreciated, otherwise I'll probably just exclude the language [12:15] If we can't find a font with the glyphs, it's probably the right answer [12:15] Otherwise there's no way to complete the install in any case [12:16] there's ttf-dzongkha [12:16] if i've mounted a remote server using the Places menu ("Connect to Server...") should those files also be accessible through the Open File dialog? [12:16] at 600KB [12:16] mjg59: does the backtrace in bug 56587 help at all? [12:16] Malone bug 56587 in usplash "[edgy] usplash segfaults" [Medium,Confirmed] http://launchpad.net/bugs/56587 [12:16] mdz: any opinions on adding the above? [12:16] We should probably find (or write) a tool that can tell us if all the characters in a po file are in the fonts [12:16] I believe it's what the d-i team is using [12:17] mdz: Not really [12:17] sabdfl: If the application understands gnome-vfs and has told the filechooser that, yes [12:17] mdz: x86emu blows up with some bioses. We only see it on amd64 because we use vm86 on x86 [12:17] sabdfl: firefox in particular won't [12:18] mdz: I've disabled the code that's likely to trigger it, since we don't use it anyway, but I'll want to track it down at some point because it hurts us in vbetool [12:18] sabdfl: oo.o should, at least when running native [12:18] mjg59: that's in your bzr? [12:18] mdz: Yup [12:19] mdz: The bzr needs a trivial fix to compile [12:19] yeah, it's t-bird that i noticed this in [12:20] gnome-vfs is solving a problem in the desktop which should probably be solved at a lower level [12:20] mjg59: ok, that will go away with a newer console-setup - it's because it's sucking the keyboard configuration out of xorg.conf, which I've stopped doing [12:20] I'll bump the dependency a bit [12:20] but the assumptions which would be broken by doing so make my head spin [12:21] Kamion: Ah, ok [12:21] mjg59: how bad is it for affected users? they see a black screen all the way until X starts? [12:21] mdz: The plan is to redevelop gnome-vfs so that it's accessible via FUSE [12:22] And then have some mechanism for providing those paths to applications that don't understand gnome-vfs natively [12:22] mjg59: it seems likely that there are all sorts of filename-handling routines which won't cope with a URL though [12:22] mdz: Yeah, that's the likely behaviour [12:22] mdz: Right, so they'll never see one [12:23] and suddenly we're all running plan 9! === lotusleaf [n=lotuslea@unaffiliated/lotusleaf] has joined #ubuntu-devel [12:23] The reason not to just do it all in FUSE is because there's no nice way to do async operations via POSIX [12:23] And you really want async file i/o when the file store is a random webdav server on a bit of string [12:23] mdz: firefox has gnome-vfs support, and some of the other distros have native dialogue patches [12:24] So gnome-vfs applications get to do things properly, and non-aware ones get to access the files anyway (just with a bit more suck) [12:25] jdub: oh, so it does. network servers don't show up in its open dialog though [12:26] mdz: I'll try testing the usplash changes here in a bit - I can force x86emu on a 32 bit system [12:26] mdz: perhaps they don't turn on that smarts in the filechooser because they can't tell if they have gnome-vfs support? not sure. [12:26] i might be unreliably bad, too, mind [12:27] mjg59: aio [12:27] mjg59: unpacking that: there is too async file operations via posix. [12:27] mjg59: can we notice if usplash segfaults and clean up after it somehow, to be robust against this problem in general? === tormod [n=tormod@80-219-112-71.dclient.hispeed.ch] has left #ubuntu-devel ["Ex-Chat"] === rodarvus [n=rodarvus@ubuntu/member/rodarvus] has joined #ubuntu-devel [12:34] mdz,tfheen: could you eyeball http://people.ubuntu.com/~cjwatson/tmp/ubiquity-1.1.24.diff for me? fixes mjg59's UK keyboard selection issue above, which I think is beta-RC [12:34] mdz: We could add a segv handler to it and chvt back to another vt === sabdfl [n=sabdfl@ubuntu/member/pdpc.silver.sabdfl] has left #ubuntu-devel [] === nictuku [n=yves@ubuntu/member/nictuku] has joined #ubuntu-devel [12:40] Keybuk: Ouch. Another reason for blacklisting nvidia is because it takes up *4MB* of kernel memory [12:40] Which is unswappable [12:43] mdz,tfheen: also, do I want to accept these changes turning Breaks back into Conflicts? [12:44] that's gnumeric, mozilla-firefox-locale-all, hplip [12:47] Setting up cupsys-driver-gutenprint (5.0.0-2ubuntu1) ... [12:47] can't close cups-genppd pipe: at /usr/sbin/cups-genppdconfig.5.0 line 418. [12:47] dpkg: error processing cupsys-driver-gutenprint (--configure): [12:47] ^-- breaks amd64 livefs build [12:47] subprocess post-installation script returned error exit status 1 [12:47] mjg59: I'm not blacklisting it until X loads it [12:47] which it doesn't [12:47] somebody else please fix [12:47] I need to go home [12:48] I've started live CD builds, which will presumably only be useful for amd64 and powerpc [12:48] er, for i386 and powerpc, I mean === dsas [n=dean@host-84-9-171-130.bulldogdsl.com] has joined #ubuntu-devel [12:49] Keybuk: Yes it does [12:49] I've just tested it [12:49] mjg59: it doesn't here [12:49] I also just tested it [12:50] I get three flickers, and then a "X could not start" [12:50] Keybuk: Did you blacklist nvidia and nvidia_legacy? [12:50] Otherwise nvidia_legacy will get loaded and X will fail because of the ABI mismatch [12:50] I just rmmod'd the running one and started X [12:50] so no module was loaded [12:51] Keybuk: I've just tested here with -10. It loaded nvidia. [12:51] Does your system use nvidia or -legacy? [12:51] nvidia [12:51] Ok [12:51] by what process does it load the module? [12:51] I've uploaded that ubiquity change; somebody please accept it if it's ok [12:52] Keybuk: I just did sudo rmmod nvidia; X [12:52] And got running X [12:52] *tests again* [12:52] nope, still not starting [12:52] I assume it uses the same mechanism that other drivers can use to trigger the loading of the DRM module [12:52] Keybuk: What's the error? [12:53] Fatal server error: [12:53] Caught signal 11. Server aborting [12:53] sweeeeeet [12:53] Ha [12:53] I think you have other problems [12:53] the unusual thing here is that there are two identical graphics cards in the machine [12:53] maybe the auto-loaded can't cope with that [12:54] I can't see why it would make a difference === CarlFK [n=carl@c-24-13-53-221.hsd1.il.comcast.net] has joined #ubuntu-devel [12:54] *shrug* [12:54] anyway I'm not going to blacklist nvidia [12:54] we can consider it for edgy+1 [12:54] Keybuk: I consider it somewhat less than ideal that we break suspend and resume for people because of a non-free driver that they may not even be using [12:55] suspend and resume are happy features that are nice if they work [12:55] Keybuk: How about we blacklist by default and then document how to disable the blacklist in the same place that we document how to enable the nvidia driver? [12:55] X starting is a mandatory feature [12:55] X starts out of the box [12:55] no, I'm not going to blacklist this by default this close to the release [12:55] let's consider this for edgy+1 [12:55] we're about to make a beta release [12:55] now is not the time to start fiddling with hardware support [12:56] Keybuk: This appears to be causing regressions [12:56] it's not causing regressions [12:56] because we loaded it by default in dapper too [12:56] Kamion: breaks into conflicts -> yes [12:56] Keybuk: And the code has changed massively since then [12:56] Kamion: I looked at your diff but I'm afraid I would need to look at quite a bit more to understand the issue [12:56] Code that we can't debug [12:57] mjg59: is it segfaulting after it has forked, or can we just check its exit code? [12:57] mdz: usplash? After it's forked. [12:57] Hm. I /think/. [12:58] mjg59: at this point suspend and resume are still mythical features that work if you're lucky === mjg59 checks === poningru [n=poningru@ip70-171-62-28.ga.at.cox.net] has joined #ubuntu-devel [12:58] Keybuk: They work on ~80% of laptop hardware [12:58] it seems better to document that blacklisting nvidia may help there, rather than "oh, if you want X to load, you'll need to remove this file" [12:58] Keybuk: "If you want X to load after you've already rewritten your config file, you also need to edit this other config file" [12:59] (Except that in my case, you don't even need to do that) [01:00] mdz: Wait a sec. Usplash doesn't fork. [01:00] mdz: So yeah, just check the return code and stick a chvt in the startup script? === BenC [n=bcollins@debian/developer/bcollins] has joined #ubuntu-devel [01:02] mjg59: that will clean up even after svgalib borkage? [01:03] mdz: Nope === theCore [n=alex@modemcable106.200-70-69.mc.videotron.ca] has joined #ubuntu-devel [01:03] mdz: That's actually slightly difficult. We could hack something with vbetool. === Zdra [n=zdra@29.19-241-81.adsl-dyn.isp.belgacom.be] has joined #ubuntu-devel [01:05] Keybuk: I've just straced my server startup - the driver checks /proc/modules (eww) and then execves modprobe === mbiebl [n=michael@dslb-084-056-234-054.pools.arcor-ip.net] has joined #ubuntu-devel [01:06] ahh [01:06] Can you do an strace -f X and see what your one does when nvidia isn't loaded? [01:06] and modprobe will therefore obey the blacklist [01:06] so won't load the nvidia module [01:06] which is why X segfaults [01:06] ? [01:07] Keybuk: if modprobe is explicitly called, the blacklist is ignored, no? [01:07] I've got it blacklisted, but this still succeeds [01:07] varies [01:07] if an init script leaks MODPROBE_OPTIONS for example [01:08] though of course, I don't have it blacklisted here, so it's not that [01:08] Keybuk: if I modprobe eth1394, I get eth1394 loaded [01:08] could be an honest race condition [01:08] Mind doing the strace and seeing what happens? [01:09] whoah, that crashed the entire machine [01:09] hard reboot too [01:09] bada-boom [01:09] mjg59: I'm *really* not comfortable about relying on this to be auto-loaded [01:09] I'm sorry [01:10] Keybuk: And I'm *really* not comfortable with us screwing over people who aren't even using the driver [01:10] then you should have spoken up at the START of the release process [01:10] not this close to the end [01:10] I only got the first bug about this last week [01:10] this will break people who upgrade [01:10] I hadn't realised it was the default [01:10] it's always been the default === _human_blip_ [n=mike@220.157.65.29] has joined #ubuntu-devel [01:10] Keybuk: It certainly hasn't. People used to need to add nvidia to /etc/modules [01:11] that was probably simply because we didn't have as reliable hardware detection as we do now [01:11] nvidia only added modaliases to their drivers fairly recently [01:11] could be that too [01:11] fairly recently is still before dapper though [01:11] so dapper released like this [01:11] which means if we suddenly blacklist the module, it appears that many people's machines will begin crashing on upgrade [01:11] I'm totally not happy about that [01:12] maybe I'm jaded, but suspend is a "would be nice" [01:12] X is a "must have" [01:12] you're jaded [01:12] totally === theCore [n=alex@modemcable106.200-70-69.mc.videotron.ca] has joined #ubuntu-devel [01:12] ok, just tested again [01:13] first time it actually did load the module [01:13] I rmmod'd, and restarted X [01:13] Keybuk: But X works fine unless the user has enabled code we have no practical way to support [01:13] and that time X crashed [01:13] I'm going to try booting with the blacklist to see whether it's an rmmod problem [01:13] Keybuk: So in your case, it could actually be that the module doesn't clean up after itself? [01:13] mjg59: so? a lot of users are enabling this code [01:13] and we're planning on enabling it by default in edgy+1, apparently [01:13] Keybuk: We *are*? [01:14] Mark *really* wants bling for edgy+1 [01:14] Mark can ask nvidia for the source, then === Keybuk blames you and your shiny compiz of doom [01:14] I believe Mark is [01:14] Excellent [01:14] When he gets it, I'll remove my objection to turning it on [01:14] no, crashed immediately on that reboot [01:15] checked, and it definitly was blacklisted and not in lsmod [01:15] Keybuk: Can you power cycle? [01:15] mjg59: sure [01:15] Rather than just warmbooting? [01:15] I'll leave it off for > 3 s :p [01:17] hard-lock up :-/ [01:17] Are these drivers made out of string, or something? [01:17] Jesus [01:17] I'll try something [01:17] I'll lift one of the graphics cards out [01:17] Keybuk: Ok [01:18] Keybuk: Of course, an alternative would be to add an init script to nvidia-glx [01:18] this feels like a race condition between the module and the X server; which is scary enough in itsself [01:18] Keybuk: That checks for the driver in use and then loads the module before X starts [01:18] Basically, we just want the module load to be conditional on information that udev doesn't have [01:18] why not use a modprobe install rule? [01:19] Sure, that would work [01:19] We could do with a small tool that told us which driver was in use [01:19] That would actually be useful for suspend/resume in any case [01:19] install nvidia grep 'Driver \"nvidia\"' /etc/X11/xorg.conf && module -Qb nvidia [01:19] or something [01:19] hmm [01:19] worked just fine with just the one in [01:19] let me try another restart [01:20] fine again [01:20] so an SLI-triggered bug [01:20] nvidia = gash [01:20] Can you give the modprobe rule a go? [01:20] better than ati. ;/ [01:21] I have hopes that amd will cause ati to open their drivers, though. [01:21] I don't have any particular objection to the modprove rule [01:21] prove? probe! :p [01:21] Keybuk: Ok, that leaves me without any objections [01:21] Keybuk: Needs to be done for nvidia_legacy as well [01:21] And possibly fglrx [01:22] can you file a bug on module-init-tools to remind me? [01:23] Will do === jinty [n=jinty@195.Red-83-58-178.dynamicIP.rima-tde.net] has joined #ubuntu-devel [01:24] with sli, are you not meant to load the driver just once ? [01:24] right [01:25] it has one Device section [01:25] because its one card, honest. [01:25] and the PCIe id of the first card [01:25] it finds the second card because it's joined to the first one === mgalvin [n=mgalvin@cpe-74-67-44-20.nycap.res.rr.com] has joined #ubuntu-devel === Arrogance [n=aks@ottawa-hs-209-217-123-254.d-ip.magma.ca] has joined #ubuntu-devel [01:26] I actually don't have it enabled on Linux [01:27] so in theory, the second graphics card sits there doing nothing === gort [n=jgbiggs@cpe-24-175-15-146.houston.res.rr.com] has joined #ubuntu-devel === Amaranth [n=travis@ubuntu/member/amaranth] has joined #ubuntu-devel === zul [n=chuck@ubuntu/member/zul] has joined #ubuntu-devel === poningru [n=poningru@ip70-171-62-28.ga.at.cox.net] has joined #ubuntu-devel === jdong_ [n=jdong@d192-24-235-141.try.wideopenwest.com] has joined #ubuntu-devel [01:41] hmm, doesnt look like we'll see final isos tonight ... === givre [n=flo@APuteaux-152-1-19-11.w82-120.abo.wanadoo.fr] has joined #ubuntu-devel === doko_ [n=doko@dslb-088-073-065-104.pools.arcor-ip.net] has joined #ubuntu-devel [01:44] mdz, do you plan to kick off alternate builds after ubiquity is done ? if so, could you also trigger edubuntu builds, so i can see if anything size wise changed right after i get up ? === keebler [n=keebler@235.119.221.216.srtnet.com] has joined #ubuntu-devel [01:46] Which one of you is Colin watson? [01:47] Kamion. === keebler Kamion Heya, would it be possible to adapt the new 256color Usplash to Dapper? [01:47] Crap [01:47] Damn dentists. [01:47] he's gone for the day [01:47] Oh. === zul [n=chuck@ubuntu/member/zul] has joined #ubuntu-devel [01:51] gone for the *night* would be better [01:52] even Colin has to sleep sometimes, to rest that planet-sized brain of his === eggauah [n=daniel@201.82.16.126] has joined #ubuntu-devel [01:56] right, and moving while releasing a beta is hard stuff === wasabi_ [n=wasabi@ubuntu/member/wasabi] has joined #ubuntu-devel === poningru [n=poningru@ip70-171-62-28.ga.at.cox.net] has joined #ubuntu-devel === shackan [n=shackan@85-18-14-13.fastres.net] has joined #ubuntu-devel [02:13] keebler, you can backport it yourself. There will not be an official 256-color usplash for dapper === robertj [n=robertj@68-117-221-30.dhcp.athn.ga.charter.com] has joined #ubuntu-devel [02:15] Seveas: Ok. [02:15] Heh, I don't suppose there is an FAQ to help me is there? [02:16] I've only been using Ubuntu for 5 months. Was a BSD junkie for the last 8 years. === Yagisan [n=Yagisan@doomsday/developer/Yagisan] has joined #ubuntu-devel [02:22] Seveas: Or a good start would be where can I aquire the usplash used in edgy? [02:22] keebler, add a deb-src line for edgy to your sources.list and apt-get source usplash === j_ack [n=rudi@p508D86D9.dip0.t-ipconnect.de] has joined #ubuntu-devel [02:23] keebler, there is the ubuntu packaging guide shipped in your docs and use the source Seveas just pointed out to build a backport [02:23] or live on the bleeding edge and use the main bzr branch [02:25] ogra: yes, I do plan to do that [02:26] mdz, thanks then i'll get some hours of sleep now and check the isos tomorrow :) === freet15 [n=freet15@221.221.160.182] has joined #ubuntu-devel === bddebian [n=bdefrees@71.224.172.103] has joined #ubuntu-devel [02:38] Heya === mgalvin_ [n=mgalvin@cpe-74-67-44-20.nycap.res.rr.com] has joined #ubuntu-devel === Hobbsee [n=Hobbsee@CPE-144-136-125-169.nsw.bigpond.net.au] has joined #ubuntu-devel === rideout [n=rideout@71-208-29-158.hlrn.qwest.net] has joined #ubuntu-devel === zul_ [n=chuck@CPE0006258ec6c1-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-devel === robertj [n=robertj@68-117-221-30.dhcp.athn.ga.charter.com] has joined #ubuntu-devel === jack_wyt [n=jack@221.221.147.52] has joined #ubuntu-devel === _human_blip_ [n=mike@220.157.65.29] has joined #ubuntu-devel [03:25] what does nautilus-cd-burner use to burn data DVD projects? [03:26] just straight growisofs -Z device ? [03:26] E: Couldn't configure pre-depend belocs-locales-bin for locales, probably a dependency cycle. [03:26] nice! === jdong_ wondering if just dragging a VIDEO_TS folder into nautilus and burning it would generate a working DVD [03:30] jdong_: probably not [03:30] DVDs are quite complex [03:30] the isofs has some odd properties, iirc [03:30] hmm [03:30] DVD players actually ignore the filesystem [03:31] instead they begin playing at particular blocks, and need padding for timing, etc. [03:31] :-/ === jdong_ hates DVD's with a passion now [03:31] ffmpeg -i $INPUT -target pal-dvd -bf 2 $FILE.mpg [03:31] mplex -f 8 -o /dev/stdout background.m2v background.m2a | spumux -v 2 spumux.xml > background.mpg [03:31] dvdauthor -o $DVD -x dvdauthor.xml [03:31] mkisofs -dvd-video -o $DVD.iso $DVD [03:31] growisofs -Z /dev/hda=$DVD.iso [03:31] -- [03:31] hmm [03:32] is my "make a video DVD" recipie [03:32] I've seen it just done with growisofs before [03:32] the -dvd-video switch to mkisofs is *very* important, it doesn't work on non-computers without it [03:32] without mkisofs first... [03:32] hmm [03:32] well, better not to risk it [03:32] growisofs will call mkisofs if you give it a directory instead of a file [03:32] so you could do growisofs -Z /dev/hda -dvd-video $DVD [03:33] but I prefer having the iso to check as an intermediate step [03:33] mmkay === jdong_ leaves the GUI appeal of nautilus === hind3nburg [n=darren@72.174.196.10] has joined #ubuntu-devel === moco [n=mojo@c211-28-37-245.sunsh1.vic.optusnet.com.au] has joined #ubuntu-devel === keebler [n=keebler@235.119.221.216.srtnet.com] has left #ubuntu-devel [] [03:37] are there any reasons behind the entry 'evolution-2.2.desktop' still in /usr/share/applications whilst it does nothing but duplicate evolution.desktop? Is that some chance the team forgot to remove it when moving up to higher version of Evo? [03:47] moco: check out the last comment of bug 57628 [03:47] Malone bug 57628 in evolution "there are multiple menu entries for evolution in a new edgy install" [Wishlist,Confirmed] http://launchpad.net/bugs/57628 [03:49] thx === jack_wyt [n=jack@221.221.163.15] has joined #ubuntu-devel === freet15_ [n=freet15@221.221.163.15] has joined #ubuntu-devel === givre [n=flo@APuteaux-152-1-88-207.w86-205.abo.wanadoo.fr] has joined #ubuntu-devel === givre [n=flo@APuteaux-152-1-88-207.w86-205.abo.wanadoo.fr] has left #ubuntu-devel [] === CarlFK [n=carl@c-24-13-53-221.hsd1.il.comcast.net] has joined #ubuntu-devel === Quinn_Storm [n=quinn@pool-71-253-28-35.pitbpa.east.verizon.net] has joined #ubuntu-devel === lastnode [n=lastnode@unaffiliated/mahangu] has joined #ubuntu-devel === cr3 [n=marc@pdpc/supporter/bronze/cr3] has joined #ubuntu-devel === zul [n=chuck@ubuntu/member/zul] has joined #ubuntu-devel === Amaranth [n=travis@ubuntu/member/amaranth] has joined #ubuntu-devel === pradeep_ [n=pradeep@59.92.50.192] has joined #ubuntu-devel [04:36] Keybuk: oddly enough, growisofs -dvd-compat doesn't imply mkisofs -dvd-video === bj__ [n=bj@ool-18bbaf15.dyn.optonline.net] has joined #ubuntu-devel [04:39] wheee...xen from ubuntu-2.6 git works http://70.29.61.171/ubuntu/Screenshot-1.png === jack_wyt_ [n=jack@221.221.162.167] has joined #ubuntu-devel === freet15__ [n=freet15@221.221.162.167] has joined #ubuntu-devel === Burgundavia_ [n=corey@ubuntu/member/burgundavia] has joined #ubuntu-devel === ogra [n=ogra@ubuntu/member/ogra] has joined #ubuntu-devel === jack_wyt__ [n=jack@221.221.163.180] has joined #ubuntu-devel === freet15___ [n=freet15@221.221.163.180] has joined #ubuntu-devel === freeflying [n=freeflyi@ubuntu/member/freeflying] has joined #ubuntu-devel === freet15___ is now known as freet15 === jack_wyt [n=jack@221.221.163.180] has joined #ubuntu-devel === crispin [n=crispin@grotto.flowerday.cx] has joined #ubuntu-devel === jldugger [n=jldugger@adsl-75-18-10-50.dsl.tpkaks.sbcglobal.net] has joined #ubuntu-devel [05:10] mdz / Keybuk: right, -dvd-compat doesn't do the trick... In fact my dvd players don't care much about it, but mkisofs's -dvd-video switch is indeed essential. [05:10] and indeed dragging some VIDEO_TS/AUDIO_TS into nautilus-cd-burner does not make a working dvd video disc [05:11] zul: do you know if anyone's working on xen wrappers for fglrx or if that'll ever happen? [05:11] jdong_: not sure for fglrx but nvidia yes.. === iwj [n=ian@xenophobe.extern.relativity.greenend.org.uk] has joined #ubuntu-devel [05:13] zul: right; I saw the work going into nvidia [05:13] oh well.... === freet15_ [n=freet15@221.221.161.53] has joined #ubuntu-devel === jack_wyt_ [n=jack@221.221.161.53] has joined #ubuntu-devel [05:17] elmo: i am on it === minghua [n=minghua@ppp-70-247-231-175.dsl.hstntx.swbell.net] has joined #ubuntu-devel [05:22] ok i see it has been done already [05:24] night === freet15__ [n=freet15@221.221.160.216] has joined #ubuntu-devel [05:26] zul: night === bj__ [n=bj@ool-18bbaf15.dyn.optonline.net] has joined #ubuntu-devel === stub [n=stub@ppp-58.8.4.98.revip2.asianet.co.th] has joined #ubuntu-devel === Fujitsu [n=profile@ubuntu/member/fujitsu] has joined #ubuntu-devel [05:40] mdz: can i declassify #61861 as non beta blocker? [05:41] bug 61861 [05:41] mdz: it's basically a bug on dapper kernel that's triggered by ruby on ppc [05:41] Malone bug 61861 in ruby1.8 "ruby1.8: fails to build on ppc under 2.6.15 kernel" [Medium,Confirmed] http://launchpad.net/bugs/61861 [05:41] fabbione: if it's able to build after elmo's changes, then yes [05:41] it was only a blocker because of the build failure [05:41] it has been fixed upgrading to .17 davis and buildd [05:41] yeah exactly [05:41] it's not an issue anymore [05:42] see RT #16838 too === jack_wyt__ [n=jack@221.221.161.6] has joined #ubuntu-devel [05:43] mdz: demoted [05:43] thanks [05:43] no problem [05:43] i guess the 2 sparc kernel bugs will wait after beta.. === Huahua [n=hua_@122.0.229.217] has joined #ubuntu-devel === freeflying [n=freeflyi@221.221.160.216] has joined #ubuntu-devel [05:45] mdz: bug #62485 is worth a ReleaseNote [05:45] Malone bug 62485 in linux-source-2.6.17 "[SPARC] - Regression - Niagara does not reboot properly" [High,Confirmed] http://launchpad.net/bugs/62485 [05:45] (beta) [05:45] basically they need a new OBP [05:45] (BIOS) [05:48] fabbione: ->Mithrandir [05:49] oky === _human_blip_ [n=mike@220.157.65.29] has joined #ubuntu-devel [06:02] has anyone noticed there's a voice print recognizer for pam (but it's defunct?) [06:02] I was thinking, cheap biometric authenticator === j_ack_ [n=rudi@p508D8540.dip0.t-ipconnect.de] has joined #ubuntu-devel [06:04] "my voice is my passport" ;) [06:06] verify me [06:06] it was speaker, password, and password length sensitive [06:06] I don't get the last two [06:07] you're speaking your password, those aren't secure variables. [06:07] The first is, until someone records you. [06:08] id just be happy if my fingerprint reader worked [06:09] there's a pam module for that too, and its not defunct, but its been sitting in debian's wnpp for over a year i believe [06:09] I'd honestly love a secure client-server signature verification mechanism. [06:09] handshake would be like [06:10] you have a GPG key, the authenticator has your UID, username, GPG key ID and fingerprint [06:11] (your username may or may not be an identifier on your gpg key; in general I fancy e-mail address but for most systems where you have to be a user of the system a particular user name is more secure) === j_ack_ [n=rudi@p508D8540.dip0.t-ipconnect.de] has joined #ubuntu-devel [06:11] authenticator says, "Who are you? Jakub? OK here's 1024 bytes of shit, sign it for me." [06:11] if it doesn't know who the hell Jakub is it just discards the signature and says "Nope don't think you fit in" [06:11] if jakub's signature is wrong it does same [06:12] if Jakub is a user; it can locate his key; the finger print matches; and Jakub's signature response is good; it allows Jakub to log in. [06:13] the 1024 bytes of junk is RANDOM and the GPG key is established through PKI or PSK depending on need so replay attacks, MiMs, and all kind of other stupid crap don't work === fabbione patiently waits for his OBP/ALOM to upgrade [06:14] jldugger: there's a GPG key PAM module but I think it works like "Verify my password really unlocks my private key," which to me looks like "open a goatse-wide security hole in the trust model of this system" [06:14] i really dont know a lot about pam or gpg === Fujitsu_ [n=fujitsu@203.23.49.35] has joined #ubuntu-devel [06:16] I really need to learn to write pam modules that don't suck [06:17] it was my intent to write a pam module that enforced hold timing of log-ins [06:17] that is, you log in on TTY1, fail, "wait 3 seconds" no screw that log in on TTY 2 and you're on by the time you time out [06:17] same with ssh, 300 concurrent connections and only 1 has to hit, 3 seconds. [06:18] what I want is that 3 seconds is applied across the board except for at the console [06:18] if you try again, we don't even check your password; we just delay and go "Nope don't know who you are" [06:19] bluefoxicy: that would piss me off no end [06:19] (when 3 months YOU spend brute forcing a 3 character long alphanumeric password, look as l33t, you will not) [06:20] HrdwrBoB: in the worst case you could DoS external connections over ssh [06:20] you're erring too far on the security vs ease of use [06:20] but you would have to know the username you're attacking [06:20] trying to log in as bauoseu is going to lock out... well nothing, that user doesn't exist on any sane system [06:20] HrdwrBoB: I said on the console it would allow it [06:21] yeah, but people often get their passwords wrong [06:21] even via ssh [06:21] sit your ass in front of the machine and log into gnome and you're good to go even if you're under attack and the attacker is failing at your user account [06:21] yes but on ssh the single ssh connection you make waits 3 seconds now [06:21] and then says "failed to authenticate" [06:22] if you can type really, really fast you can start another ssh session and get logged in within that 3 seconds [06:22] more usefully, an attacker can make 30 ssh connections at once and launch attacks in parallel; and if he can do 30 at once x 10 times a second, that's 300 attacks a second [06:22] 300/s is SFA [06:23] So, you basically just want PAM locking. That sounds trivial to implement. [06:23] it's highly unlikely you're actually going to be dumb and spend hours trying to ssh in [06:23] when the solution is "hold on for 3 seconds" [06:23] bluefoxicy: I doubt you can do parallel SSH brute forcing with decent efficiency [06:23] especially since you have to willfully kill/restart the ssh client or use another terminal to ssh in real fast to make the window (as I said, ssh does make you wait when you get it wrong) [06:24] jdong_: It gets attempted on my machines several times weekly. [06:24] jdong_: as do I; but I prefer "I can guarantee" to "I doubt" [06:24] jdong_: To the point of saturating my DSL sometimes. [06:24] infinity: wow... that's phenomenal indeed... my ssh has remained relatively unattacked [06:24] infinity: yeah, I tried based on pam-audit and failed hard === rgould [n=rgould@S0106000f3d657273.gv.shawcable.net] has joined #ubuntu-devel === freet15_ [n=freet15@221.221.161.6] has joined #ubuntu-devel [06:26] bluefoxicy: Check for lockfile, spin until lock released. Stack pam_foreground appropriately to bypass your pam_locking for foreground console users. [06:26] umm, why doesn't the linux-server metapackage pull in the latest kernel? [06:27] yeah [06:27] jdong_: Does for me. [06:28] oh hah [06:28] silly me === jdong_ reboots server === Gman is now known as GmanAFK === jdong_ [n=jdong@d192-24-235-141.try.wideopenwest.com] has joined #ubuntu-devel [06:34] ugh. Malone 17297 is annoying. [06:34] Malone bug 17297 in synaptic "Orphan Search" [Wishlist,Needs info] http://launchpad.net/bugs/17297 [06:36] I'm beginning to think I'm actually the only person in the universe that ever actually cleans old packages off his disk [06:37] (no, the auto-removable packages stuff is NOT doing it for me, I've cleaned out all its suggestions already) [06:37] bluefoxicy: I just reinstall ;-) [06:37] LaserJock: I try to avoid rebooting [06:38] haha [06:38] LaserJock: speaking of reinstall did I ever file a bug or spec out "Wipe everything on target / except /home during reboot"? [06:38] since everyone absolutely refuses to isolate /home from / [06:38] and some people have 350 gigabytes of /home [06:38] seems like you did [06:38] can't remember exactly though [06:39] "oh just burn it to {CD,DVD}" I HAVE A HUNDRED GIGS. TWENTY FIVE DVDS. The stock answers I get are totally annoying. [06:39] hmm, I just back it up since I want to wipe the drive anyway [06:40]